Lightpanda 0.4.1 drops CORS, 140x faster selectors

// 1h agoPRODUCT UPDATE

Lightpanda 0.4.1 drops CORS, 140x faster selectors

Lightpanda has released version 0.4.1 of its Zig-based headless browser for AI agents, adding experimental CORS enforcement and roughly 140x faster nth-* CSS selector matching. The update also introduces PCRE2 regex filtering for native ad blocking, new strip modes for cleaner web extraction, and HTTP WebDriver session management.

// ANALYSIS

Lightpanda is quietly building the strongest alternative to Chrome Headless for autonomous web agents by trading graphical rendering bloat for raw execution speed and lower memory usage.

–Dramatically lower DOM parsing latency: The ~140x speedup on nth-* selectors and optimized class lookups removes serious friction when crawling dynamic, deeply nested single-page applications.
–Pre-filtering token bloat: Upgraded native ad blocking with PCRE2 regex and new strip mode options allow pipelines to strip visual noise before sending HTML to LLM context windows.
–Bridging the web compatibility gap: Adding opt-in CORS support and WebDriver session management provides the compliance required by modern web apps without sacrificing Lightpanda's lightweight runtime.
–Native agent ergonomics: Support for full mouse pointer event dispatching and persistent HTTP connections ensures AI agent interactions behave reliably across multi-step browser tasks.

// TAGS

lightpandaopen-sourcedevtoolautomationweb-agentagent

DISCOVERED

1h ago

2026-09-16

PUBLISHED

1h ago

2026-09-16

RELEVANCE

8/ 10

AUTHOR

lightpanda_io

// KEEP READING

More AI developer news from the feed

EXPLORE FULL FEED

VIDEO1h ago

OnSpace AI builds full-stack apps from prompts

OnSpace AI is a no-code development platform that allows users to create, test, and deploy full-stack web and native mobile applications through conversational natural language prompts. The service removes typical backend and DevOps overhead by providing managed database storage, turnkey authentication, and built-in access to large language models without requiring users to configure custom API keys. It includes an automated runtime error scanner to identify and resolve code issues on the fly, real-time interactive previews, and streamlined publishing workflows directly targeting the Apple App Store and Google Play Store.

UPDATE1h ago

Mercury Agent adds multi-model cybersecurity recon

Mercury has announced that its upcoming release will incorporate native cybersecurity capabilities, kicking off with a dedicated `/recon` command designed for authorized, white-hat security operations. The tool uses Mercury to orchestrate workflows across multiple AI models while maintaining a unified context and consistent operational flow, marking the initial release in a planned suite of security-focused agent features.

RESEARCH3h ago

Salesforce drops Koa enterprise agent model

Salesforce published research detailing Koa, a 120-billion-parameter enterprise language model optimized for agentic tool use and multi-step CRM workflows within its Agentforce platform. Built by post-training NVIDIA's Nemotron-3-Super-120B with GRPO, Koa uses a simulation-to-reward pipeline based on declarative Agent Script configurations to match frontier models on enterprise benchmarks without customer data.

Lightpanda 0.4.1 drops CORS, 140x faster selectors